4 Wheel Motored Vehicle ”UOT Electric March II”

-Experimental EV for Novel Motion Control Studies-

,Shin-ichiro SakaiAssistant ResearcherThe Institute of Space andAstronautical ScienceSagamihara, Kanagawa 229-8510

Takahiro Okano, Tai Chien HwaCandidates for MEngThe University of TokyoBunkyo, Tokyo 113-8656

Toshiyuki UchidaTechnical StaffThe University of TokyoBunkyo, Tokyo 113-8656

Yoichi HoriProfessorThe University of TokyoBunkyo, Tokyo 113-8656

KEYWORDS

Electric vehicles, Motion Control, Anti-lock Braking System, Direct Yaw Moment Control.

ABSTRACT

“UOT(University of Tokyo) Electric March-II” is our novel experimental EV. It is 4-wheel mo-tored EV: every wheel has its own driving motor. Each motor can be fully driven independently.This EV is for intensive studies of “motion control of electric vehicle”, since it will be an impor-tant research issues. EV has great advantage on control performance: fast and accurate motortorque generation. Therefore, advanced methods can be applied for wheel skid prevention orroad condition estimation. With small motors like in-wheel motors, even the lateral motioncontrol is available. This paper first introduces “UOT Electric March II”, then describes suchresearch topics which were and will be demonstrated with our new EV.

1 INTRODUCTION

Recently, electric vehicles (EVs) are intensively developed. With improvement of motors andbatteries, some pure EVs (PEVs) with only secondary batteries have already achieved enoughperformance. Hybrid EVs (HEVs), like Toyota Prius, are going up to the commercial products.Fuel cell EVs (FCEVs) will possibly be major vehicles in this 21st century. The background ofthis developments is energy and environmental problems, thus main concern over EVs is energyefficiency and environmental impacts. However, another important advantage exists, which isnot recognized well yet. It is controllability of electric motors.

From the viewpoint of electrical and control engineering, EVs have evident advantages overinternal combustion engine vehicles (ICVs). These advantages can be summarized as:

1. Torque generation of electric motor is very quick and accurate, for bothaccelerating and decelerating.

This should be the essential advantage. ABS (antilock brake system) and TCS (tractioncontrol system) should be integrated into “total TCS”, since a motor can both accelerateor decelerate the wheel. Its performance should be advanced one, if we can fully utilizethe fast torque response of motor [1].

2. Motor torque is easily comprehensible.

There exists little uncertainty in driving or braking torque inputted by motor, compared tothat of combustion engine or hydraulic brake. Therefore, simple “driving force observer”can achieve a real-time observation of driving/braking force between the tire and roadsurface [2] [3]. This second advantage will contribute a great deal to several applicationslike road condition estimation.

3. Motor can be attached to each wheel.

With small motors like in-wheel motors (Fig. 1), even the anti-directional torque generationis possible on left and right wheels. In automobile engineering, such approach is knownas DYC (direct yaw moment control) [4] [5]. Distributed motor will possibly enhance itsperformance.

Definitely, these indicate the novel approach for vehicle motion control in EVs. Automobileengineers recently assume that active vehicle control is the important technique. It is directlyconnected with safety, or human life. We, electric engineers, will be able to contribute a lot tothis novel and important theme: advanced motion control of EVs.

“UOT Electric March II” is a novel experimental EV, which has been constructed for thesestudies in 2001. It is 4 wheel motored EV: every wheel has its own driving motor(Fig. 1).Computers for motion control, sensors like fiber-gyro type yaw rate sensor, and inverter unitswhich generates the commanded torque are also equipped. Thus it is an ideal vehicle for bothlongitudinal and lateral motion studies.

In the following parts, section 2 introduces this laboratory-made EV. Subsequent sectionsmention the research topics which this EV is constructed for. Feedback based approach withfast motor response is our basic strategies(Fig. 2). Such minor controller can change the plantdynamics and enhance the stability on slippery road surface. Longitudinal stability is discussedin section 3, and section 4 concerns the lateral motion stability enhancement with this minorcontroller. Lateral stability is studied not only with simulations, but also experimental resultswith “UOT Electric March II.” Other research issues, which are planed to be demonstrated withthis EV, are shortly introduced in section 5. Section 6 concludes this paper.

In-Wheel Motors

Inverter unitMain Batteries(18 Units, 216[V])

Motion Sensors(Yaw Rate Sensor, Acceleration Sensors)

Electric Power Steering

Inverter Unit

PC for Control

12V Supply

DC-DC Converter

Motors

Electrical Wire Harnesses

Figure 1: In-wheel motor (left) and our new EVwith four in-wheel motors. (right)

outer loop of chassis control,based on measured yaw rate and/or observed slip angle, etc.

(2 o

r 4

mot

ors)

fast minor loopsfor each motor

Motor or Wheel Vel. Vw

Fm

Yaw Moment reference Mz

Motor torque reference Fm

*

Driving/Braking Force Fd

+

+

+

-

Skid Detector

MFC

Dri

ving

/Bra

king

For

ce D

istr

ibut

or

Vehicle

MotionController (DYC)Vehicle

δ fSteering Angle

Electric Motor

Figure 2: Our basic idea: total systemwith minor feedback loops

2 EXPERIMENTAL EV “UOT ELECTRIC MARCH II”

“UOT Electric March II” is our novel EV for experimental discussions. This EV can be char-acterized by its original motor configuration: 4 independent motors. Each wheel has their owndriving motor, therefore, driven wheels can be independently controlled. Regenerative brakingis also available in this vehicle. We ourselves designed and built up this vehicle, based on theconventional ICV “Nissan March”.

The specification of this EV focuses on the motion control experiments. It has adequatedevices for experiments: on-board PCs and several sensors like fiber-optic gyro. Motion controllerconstructed in the PC outputs the torque commands, and inverter units generate the torquesof these values. This precise torque generation is achieved by motor current controller in theinverter unit. On the other hand, the distance for one-charge, energy efficiency, driving comfortor other features are excluded from discussions. Table 1 summarizes the key specifications of“UOT Electric March II”.

Figure 3: “UOT Electric March II” running at about 100 [km/h].

Table 1: Spec. of “UOT Electric March II”.

Drivetrain 4 PM MotorsMax. Power(20 sec.) 36 [kW] (48.3[HP])∗

Max. Torque 77∗ [Nm]Gear Ratio 5.0Battery Lead AcidWeight 14.0 [kg](for 1 unit)

Total Voltage 228 [V] (with 19 units)Base Chassis Nissan March K11Wheel Base 2360 [m]

Wheel Tread F/R 1365/1325 [m]Total Weight 1400 [kg]

Wheel Inertia∗∗ 8.2 [kg]∗∗∗Wheel Radius 0.28 [m]Controller

CPU MMX Pentium 233[MHz]Rotary Encoder 3600 [ppr]∗∗∗

Gyro Sensor Fiber Optical Type

* ... for only one motor.** ... mass equivalent.*** ... affected by gear ratio.

Figure 4: Photo of turning experiments with 4W-motored EV “UOT Electric March II” (Fig. 12-13).

3 WHEEL VELOCITY CONTROLLER FOR SKID PREVENTION

In this section, the wheel velocity controller for skid prevention is discussed. The starting pointof this idea is to utilize the knowledge on motion control, which is based on the motor control.In general, the feedback controller can change the dynamics of plant, or we can re-design theplant dynamics. For example, the plant can be insensitive against disturbance if appropriatefeedback controller is applied. Such feedback controller requires fast response of actuator, andit is available in EVs. So, how we should design the controller or plant dynamics for skidprevention? This is the main topics in this section.

3.1 SLIP PHENOMENA AND LINEAR SLIP MODEL

Ordinary, slip ratio λ is used to evaluate the “slip”. Slip ratio λ is defined as,

λ =

Vw − V

Vw

: for accelerating wheel,

Vw − V

V: for decelerating wheel,

(1)

where V is the vehicle chassis velocity. Vw is the velocity equivalent value of wheel velocity,Vw = rω, where r, ω are the wheel radius and wheel rotating velocity, respectively.

With simple one wheel model (Fig. 5), the motion equations of wheel and chassis can beobtained as

MwdVw

dt= Fm − Fd(λ), (2)

MdV

dt= Fd(λ), (3)

if air resistance on chassis and rotating resistance on wheel are both negligible. M and Mw

are the vehicle weight and the mass equivalent value of wheel inertia, respectively. Fm is theforce equivalent value of accelerating/decelerating torque, generated by engine, hydraulic brakesystem or motor. Fd is the driving/braking force between the wheel and the road surface. ThisFd has nonlinear dependence on the slip ratio λ, such as in Fig. 6 1.

For the controller design process, linear skid model is derived from (1)-(3) and Fd(λ) in Fig. 6.Nonlinearity exists in Fd(λ) or µ − λ curve, therefore, perturbation equation for Fd(λ),

∆Fd = N∆µ = Na∆λ (4)

= − 1

Vw0

∆V +V0

V 2w0

∆Vw (5)

is used here. The parameter a is the gradient of µ − λ curve,

a =∂µ

∂λ

∣∣∣∣(V0,Vw0)

. (6)

Vw0, V0 are the wheel velocity and chassis velocity at the operational point, respectively. With(1)-(3) and (5), the transfer function from motor torque Fm to the wheel velocity Vw is

P (s) =∆Vw

∆Fm=

1

(Mw + M(1 − λ0))s

τws + 1

τas + 1, (7)

1µ = Fd/N , where N is the normal force on the wheel.

Vw

Fd

M

V

Mw

FmN

Figure 5: One wheel model.

0

0.2

0.4

0.6

0.8

1

Dry Asphalt

Icy or Snowy Road

Wet Asphalt

0 0.2 0.4 0.6 0.8Slip Ratio

Dri

ving

For

ce, S

ide

For

ce (

norm

aliz

ed)

Side Force (tire slip angle = 4.0 [deg])

1

Figure 6: Typical µ−λ curve.

-+

+ ++

-

d d: disturbance�: sensor noise

Fm Vw

Kp

+

-

Fm*

Padh(s)

s�s+1Mw

Inverse model + LPF = HPF

P(s)

�

Figure 7: Proposed feedback controller.

where

τa =MwVw0

aN

M

M(1 − λ0) + Mw

, τw =MVw0

aN. (8)

λ0 is a slip ratio at the same operational point (V0, Vw0).From (7)-(8), the most simple models Padh(s) (for adhesive wheel) and Pskid(s) (for completely

skidding wheel) are

Padh(s) =1

M + Mw

1

s, Pskid(s) =

1

Mw

1

s. (9)

3.2 CONTROLLER DESIGN

Therefore, One dominant phenomenon in the wheel skidding is the rapid change of wheel rotatingvelocity. With wheel skidding during the acceleration, the wheel velocity rapidly increases, andduring the deceleration it rapidly drops due to the wheel lock. Eq. (9) describe that sudden dropof wheel equivalent inertia causes this rapid change of wheel velocity. Based on this viewpoint,we design the feedback controller of Fig. 7 [1]. This controller can suppress such sudden drop ofinertia as shown in Fig. 8.

Fig. 8 is the bode diagram of Vw/F ∗m. Left graph plots Vw/F ∗

m for wheel without controller,i.e., plots Padh(s) and Pskid(s). If the controller of Fig. 7 is applied, these transfer functions arechanged into the ones in the right graph. These figures clearly indicate that the dynamics ofskidding wheel comes to be almost same as that of adhesive wheel, the “heavy” wheel. Thewheel with proposed controller is insensitive for slip phenomena.

-140

-120

-100

-80

-60

-40

-20

10 10 10 10 10Tra

nsfe

r F

ncti

on V

w/F

m*

[db]

Tra

nsfe

r F

ncti

on V

w/F

m*

[db]

Skidding Wheel, w/o Control

Adhesive Wheel, w/o Control

Without Control

Frequency [rad/sec]-1 0 1 2 3

-140

-120

-100

-80

-60

-40

-20

10 10 10 10 10Frequency [rad/sec]

Skidding Wheel, w/o Control

Adhesive Wheel, w/o Control

Adhesive Wheel, with Control

Skidding Wheel, with Control

With Control

-1 0 1 2 3

Figure 8: Bode diagram of Vw/F ∗m. Kp = Kp

∗ = M+MwMw

. τ = 0.1[sec].

3.3 EXPERIMENTAL RESULTS

Experiments were carried out to confirm the proposed method. These experiments were carriedout with “UOT Electric March-I”, which is our another laboratory-made EV (Fig. 9) constructedin 1997 [1]. To examine the effect of wheel velocity control for skid avoidance, slippery low µroad is required. We put the aluminum plates of 14[m] length on the asphalt, and spread wateron these plates. The peak µ of this test road is about 0.5. This value was estimated based onsome other experimental results [3].

Fig. 10 shows the time responses of slip ratio. In these experiments, vehicle accelerated onthe slippery test road, with lineally increasing motor torque. Without control, the slip ratiorapidly increases. On the contrary, the increase of slip ratio is relatively slow with proposedcontroller. As mentioned above, the wheel equivalent inertia during the wheel skidding comes tobe “heavy” with wheel velocity control, thus the rapid increase of slip ratio can be suppressed.

Note that this method cannot be a complete skid prevention controller by itself. Proposedcontroller suppressed the rapid growth of slip ratio, however, the slip ratio finally exceeded thestable limit (Fig. 10). Therefore, we suggest this method as a minor-loop controller, to improveother method like conventional ABS or skid detection technique with EV [7].

(1800 ppr)

acceleration command

motor current output command

CurrentSensor

Battery

rear tire velocity (120ppr)

CounterBoard

1 Quadrant Chopper

DC WoundedMotor

Note- PC

A/D, D/A converters

front wheel velocity

Figure 9: “UOT Electric March-I”.

Kp=0

Kp=1

Kp=5

Kp=10

Kp=20

Time [s]

Slip

Rat

io

Optimal Slip Ratio

0

0.2

0.4

0.6

0.8

1 2 3 4 5 6

Figure 10: Effect of wheel velocity control for skid pre-vention with τ=0.1[s] (Experimental results.) K∗

p is 4.52for this vehicle.

4 LATERAL MOTION STABILIZATION WITH MOTORCONTROL

4.1 CONCEPT AND SIMULATION STUDIES

In the previous section, wheel velocity feedback method was discussed. With this method, wheelequivalently has heavy inertia during slip. This suppresses the rapid increase of slip ratio. Then,what will happen if we apply such feedback loop for every wheel of turning vehicle on slipperyroad ?

As commonly known, the vehicle lateral motion can be sometimes unstable. This instabil-ity occurs in such situation as rapid braking during the turning, especially with slippery road

condition with snowy or rainy weather. Here we assume that one small motor is attached onevery wheel of target EV. In-wheel motor is a typical example (Fig. 1). With such motors, thewheel velocities can be controlled independently. Our simulation results (Fig. 11) show that thisminor loops can enhance the vehicle’s lateral stability [6]. Chassis’s 3-DOF nonlinear motion,four wheel’s rotation and dynamic load distribution are calculated in these simulations.

In these simulations, the vehicle starts running on the slippery road (µpeak = 0.5), turningleft with steering angle δf = 3 [deg]. Then at 5.0 [sec], the driver inputs rapid braking torqueFm = −1100[N] on each wheel. This torque exceeds the tire performance. Therefore, the wheelskid occurs and the chassis starts the spin motion, although the driver stops braking at 9.0 [s].This wheel skidding is serious at rear-left wheel especially, since the center-of-gravity is shiftedand the load distribution varied.

On the contrary, if the wheel velocity controller is applied independently for each wheel, suchdangerous spin motion is prevented. The rear-left wheel’s torque is most reduced automatically.Note that this method uses only wheel velocities as feedback signals, therefore, differs consider-ably from conventional methods like DYC [4] [8]. The autonomous stabilization of each wheel,which is achieved with wheel velocity feedback, enhances the stability of vehicle lateral motionon slippery road. This effect is demonstrated in the next subsection.

0 20 40 60 80 100 120 140 160 180 200

0

20

40

60

80

100

0.0 [s] 1.0 [s] 2.0 [s] 3.0 [s]

4.0 [s]

5.0 [s] 6.0 [s]

7.0 [s]

8.0 [s]

9.0 [s]

10.0 [s]

Without Control.

With Weak Feedback Control.(MFC, Gain Kp = 1)

With Feedback Control.(MFC, Gain Kp = 5)

Steering input (from 0 [deg] to 3 [deg])

Rapid Brake

Spin!

Slight Spin Motion

Stable Cornering

Vehicle Size ... x4

[m]

Rapid Brake (during turning the curve, on slippery road)

Figure 11: Stabilizing effect with “controlled four wheels” is visualized with vehicle’s trajectory [6].

4.2 BASIC EXPERIMENTAL RESULTS WITH “UOT ELECTRIC MARCH II”

Then the results of first experiments using “UOT Electric March II” is introduced here. Inthese experiments, “UOT Electric March II” was turning on the slippery test road, so-calledskid pad (Fig. 4). At first, it was making steady turning in the clockwise direction. Turningradius and chassis velocity were about 25-30 [m] and 40[km/h], respectively. These values wereclosed to the unstable region. At 0 [s], acceleration torque of 1000 [N] was applied for rear twomotors. Without any feedback control, this excessive acceleration causes the unstable vehiclemotion. Fig. 12 shows this unstable vehicle motion 2. The rear-right or rear-inside wheel startedskidding seriously. Then yaw rate γ unstably grew as shown in the upper-right graph of Fig. 12.It indicates the spin motion. Vehicle was completely out of control and at 2[sec], experimentwas terminated for safety reasons.

On the contrary, such dangerous motion could be prevented with minor feedback of wheelvelocity. Fig. 13 and Fig. 14 shows this effect clearly. Note that controllers on rear-left and

2Chassis velocities in Figs. 12 and 13 are the mean values of trailing front wheels.

rear-right wheels are the same and independent ones. Each controller only requires the value ofeach wheel’s velocity, thus it is not “connected” with each other in any meanings. Consequently,it can be said that autonomous stabilization of each driven wheel was achieved, and it enhancedthe vehicle lateral stability. This indicates the validity of simulations in the previous section.

One of the remaining problems is the high-frequency oscillation induced at the rear wheels.We suppose that it depends on the design of controller. The cut-off frequency τ in the proposedcontroller (Fig. 7) may have the important influence on this oscillation, however, such discussionsmust wait for the next experiments.

0 1 2 3 4 50

5

10

15

20

Time [s]0 1 2 3 4 5-40

-30

-20

-10

Time [s]

0 1 2 3 4 50

5

10

15

20

Time [s]0 1 2 3 4 5

Time [s]

0 1 2 3 4 50

5

10

15

20

Time [s]0 1 2 3 4 5

Time [s]

0 1 2 3 4 50

5

10

15

20

Time [s]0 1 2 3 4 5

Time [s]

0 1 2 3 4 50

5

10

15

20

Time [s]0 1 2 3 4 5

Time [s]

-2000

0

2000

-2000

0

2000

-2000

0

2000

-2000

0

2000

V [

m/s

]V

w [m

/s]

Cha

ssis

Vel

.W

heel

Vel

.V

w [m

/s]

Whe

el V

el.

Vw

[m/s

]W

heel

Vel

.V

w [m

/s]

Whe

el V

el.

γ [d

eg/s

]M

otor

Tor

que

[N]

Mot

or T

orqu

e [N

]M

otor

Tor

que

[N]

Mot

or T

orqu

e [N

]Y

aw R

ate

Front Right

Rear Right

Front Left

Rear Left

Rear Right

Front Left

Rear Left

Front Right

Figure 12: Unstable turning with sudden acceler-ation torque on rear wheels. Vehicle made steadyturning before torque inputs.

0 1 2 3 4 50

5

10

15

20

Time [s]

V [

m/s

]

0 1 2 3 4 5-40

-30

-20

-10

Time [s]

γ [d

eg/s

]

0 1 2 3 4 50

5

10

15

20

Time [s]

Vw

[m/s

]

0 1 2 3 4 5Time [s]

0 1 2 3 4 50

5

10

15

20

Time [s]0 1 2 3 4 5

Time [s]

0 1 2 3 4 50

5

10

15

20

Time [s]0 1 2 3 4 5

Time [s]

0 1 2 3 4 50

5

10

15

20

Time [s]0 1 2 3 4 5

Time [s]

Mot

or T

orqu

e [N

]M

otor

Tor

que

[N]

Mot

or T

orqu

e [N

]M

otor

Tor

que

[N]

Yaw

Rat

e

Cha

ssis

Vel

.W

heel

Vel

.V

w [m

/s]

Whe

el V

el.

Vw

[m/s

]W

heel

Vel

.V

w [m

/s]

Whe

el V

el.

-2000

0

2000

-2000

0

2000

-2000

0

2000

-2000

0

2000

Front Right

Rear Right

Front Left

Rear Left

Rear Right

Front Left

Rear Left

Front Right

Figure 13: Stabilizing effect of wheel velocityfeedback. Proposed controller of Fig. 7 was ap-plied on both rear wheels.

-50

-40

-30

-20

-10

Time [s]

Yaw

Rat

e γ

[deg

/s]

0

10

20

30

Time [s]

Whe

el V

el. V

w [m

/s]

(Rea

r R

ight

)

0 1 2 3 4 50 1 2 3 4 5

(b) Kp = 0.05 Kp*(a) Kp = 0

(c) Kp = 0.5 Kp*

(b) Kp = 0.05 Kp*(a) Kp = 0

(c) Kp = 0.5 Kp*

Figure 14: Comparison of vehicle motion: (a) without feedback controller, (b) with weak feedbackcontroller, (c) and adequate feedback controller. K∗

p is 45.2 for this vehicle. τ in the controller was0.1 [s].

5 OTHER STUDIES WITH “UOT ELECTRIC MARCH II”

We have studies several motion control issues around EV. Experimental evaluation with “UOTElectric March II” will clarify their validity. Some of these topics are shortly introduced here.

5.1 ROAD CONDITION ESTIMATION [2]

The road surface condition is quite useful information for the motion controller. This informationwill enhance the performance of ABS or TCS. DYC can avoid unintended excessive torque, andcan keep the control torque below the notified road surface limit. Therefore, the road conditionestimation is intensively studied for conventional vehicles [9]. The accurate value of wheel inputtorque will contribute a great deal to the the practical and precise estimation. It is available withEV or electric motor, but no so easy with ICV or combustion engine. We have proposed advancedroad condition estimator for EV, which estimates the µpeak value during adhesive driving [2].Basically it was confirmed with experiments, however, further studies should be carried out.

5.2 HYBRID ABS [6]

Generally speaking, HEV has only small motor for torque assist. Thus the regenerative brakingmust cooperate with hydraulic braking system(Fig. 15). This cooperation is designed only forthe energy efficiency, not for the wheel skid prevention. We have proposed “Hybrid ABS(H-ABS)”, which is the cooperative ABS with electric and hydraulic torque. The point is that,HEV’s motor has relatively small but rapid torque output, and hydraulic braking system haslarge but slow torque generation.

Currently, two approaches are discussed. First approach is the “plug-in H-ABS”. Motorcontroller is just added to the normal hydraulic ABS, without changing the original ABS con-troller (Fig. 16). Generally, the ABS controller is on-off type controller. The plug-in feedbackcontroller prevents the rapid change of wheel velocity with feedback, accordingly compensatesthe high frequency dynamics of original ABS.

Hydraric Braking Systemwith ABS

Vw

Fmotor

FABS

Vehicle DynamicsRegenerative Braking with Motor

Regenerative Braking Torque command

Hydraulic Braking Torque command

FABS*

Fmotor*

Fbrake

Figure 15: Conventional Regenerative Brak-ing. ABS actuator is only hydraulic one.

Regenerative Braking Controllerto Improve ABS performance

Vw

Fmotor

FABS

CFF Vehicle Dynamics

Pn

Q

Regenerative Braking Torque command

Hydraulic Braking Torque command

FABS*

Fmotor* Fbrake

Feedforward Compensator

Hydraulic Braking Systemwith ABS

Feedback Compensator

Figure 16: “H-ABS”, cooperative ABS withboth electric and hydraulic torque.

The applied feedback controller in Fig. 16is very similar to the one in section 3.This controller prevents the sudden changeof wheel inertia or wheel velocity, thus thewheel velocity oscillation can be suppressed.Simulation results (Fig 17) shows this effecttypically. Accordingly, the braking distancecan be shorten. However, this is just the sim-ple simulation results and farther experimen-tal studies should be carried out.

This “plug-in” type’s advantages are, (a)easy to apply and (b) only concerning withwheel velocity, and not using chassis veloc-ity for feedback signals. Another approachis, of course, to design both regenerative andhydraulic braking controller. Controller de-sign with frequency-division seems to be ef-fective with our basic simulations. Experi-mental study of this method is also planedwith “UOT Electric March-II”.

-4000

-2000

0

2000

0 2 4 6 8 10 12

Fbr

ake

[N] Fbrake = Fmotor+FABS

-4000

-2000

0

2000

0 2 4 6 8 10 12

Fbr

ake

[N]

Fbrake = Fmotor+FABS

-4000

-2000

0

2000

0 2 4 6 8 10 12FA

BS,

Fm

otor

[N

]

Fmotor

FABS

-4000

-2000

0

2000

0 2 4 6 8 10 12FA

BS,

Fm

otor

[N

]

Fmotor

FABS

Time [s]Time [s]

0

20

40

60

80

0 2 4 6 8 10 12

V, V

w [

km/h

]

VVw

Time [s]0

20

40

60

80

0 2 4 6 8 10 12

V, V

w [

km/h

]

VVw

-0.4

-0.2

0

0 2 4 6 8 10 12

Slip

Rat

io

Time [s]

with Motor Cont.

w/o Motor Cont.

0

20

40

60

80

0 2 4 6 8 10 12

V [

km/h

]

Time [s]

Braking Distance

Ratio : 0.81

w/o Motor Cont.: 98.6 [m]with Motor Cont.: 79.9 [m]

with Motor Cont.w/o Motor Cont.

without Motor Control

with Motor Control

chassis /wheel velocitiestotal braking torquehydraulic / electric torque

chassis /wheel velocitiestotal braking torquehydraulic / electric torque

Figure 17: Effect of plug-in H-ABS. Upper row shows the data of simulated conventional ABS. Simple on-offABS logic causes oscillation. Applied controller with electric motor can suppress this oscillation with compensat-ing the high frequency dynamics(middle row). Consequently, slip ratio oscillation is relatively small and brakingdistance is shorten(lower row) with proposed methods.

6 CONCLUSION

In this paper, the novel experimental EV “UOT Electric March II” is introduced. This 4 wheelmotored EV is expected to play an important roll in the motion control studies. The individualresearch topics, which will be demonstrated with this EV, are also mentioned. Please refereach paper listed below, since the details of these topics are omitted here. In addition, fartherinformation of our EV is available at: www.hori.t.u-tokyo.ac.jp/997/sakai.

Again the point is the electric motor’s advantage: quick and accurate torque generation anddistributed torques. Currently main concerning around EV is energy efficiency. In the nearfuture, control issue will be another major topics. We will continue our effort toward that day.

References[1] Y. Hori, Y. Toyoda, and Y. Tsuruoka. Traction control of electric vehicle: Basic experimental results using

the test EV “UOT Electric March”. IEEE Trans. Ind. Applicat., Vol. 34, No. 5, pp. 1131–1138, 1998.

[2] Hideo Sado, Shin-ichiro Sakai, and Yoichi Hori. Road condition estimation for traction control in electricvehicle. In Proc. The 1999 IEEE International Symposium on Industrial Electronics, pp. 973–978, 1999.

[3] S. Sakai, H. Sado, and Y. Hori. Novel skid avoidance method without vehicle chassis speed for electric vehicle.In Proc. International Power Electronics Conference (IPEC), Vol. 4, pp. 1979–1984, 2000.

[4] Yasuji Shibahata, et al. The improvement of vehicle maneuverability by direct yaw moment control. In Proc.International Symposium on Advanced Vehicle Control (AVEC) ’92, No. 923081, 1992.

[5] Sumio Motoyama, et al. Effect of traction force distribution control on vehicle dynamics. In Proc. InternationalSymposium on Advanced Vehicle Control (AVEC) ’92, No. 923080, 1992.

[6] Shin-ichiro Sakai and Yoichi. Hori. Advanced vehicle motion control of electric vehicle based on the fastmotor torque response. In Proc. 5th AVEC, pp. 729–736, Michigan, USA, 2000.

[7] Shin-ichiro Sakai, Hideo Sado, and Yoichi Hori. Motion control in an electric vehicle with 4 independentlydriven in-wheel motors. IEEE Trans. on Mechatronics, Vol. 4, No. 1, pp. 9–16, 1999.

[8] Y. Furukawa and M. Abe. Direct yaw moment control with estimating side-slip angle by using on-board-tire-model. In Proc. 4th International Symposium on Advanced Vehicle Control, pp. 431–436, Nagoya, 1998.

[9] Fredrik Gustafsson. Slip-based tire-road friction estimation. Automatica, Vol. 33, No. 6, pp. 1087–1099, 1997.